Search papers, labs, and topics across Lattice.
3
2
6
7
Forget painstakingly crafting prompts by hand – RPT uses LLM function calling to automate prompt optimization, achieving up to 12.9 point improvements on reasoning tasks and better confidence calibration.
Forget querying a single database – Blue's Data Intelligence Layer treats LLMs, the web, and even the user as queryable "databases" to answer complex, real-world questions.
LLMs struggle to track state across multiple tool-use steps, but a surprisingly simple fix—restating prior variable values—yields substantial performance gains.
